Hack31.Clean Music Metadata at the Command Line

Hack 31. Clean Music Metadata at the Command Line

Use Lucene-based tagging in MusicBrainz to automagically scan and tag all your MP3s from the command line.

Most people never understand the importance of clean metadata (e.g., ID3 tags) in their music until they shove their haphazard music collection into an iPod. If you have no tags or inconsistent tags in your music collection, it's nearly impossible to really enjoy your shiny new iPod. To help out with this dilemma, MusicBrainz has been working on creating intelligent tagging applications that take the pain out of fixing your music metadata (for an introduction to MusicBrainz, see [Hack #30]).

MusicBrainz offers a full-blown GUI application code-named Picard [Hack #32] that provides a visual approach to tagging your music. Musicbrainz also provides pimpmytunes for people who prefer to work on the command line. MusicBrainz tagging applications use the Lucene engine (http://lucene.apache.org) behind the scenes to identify tracks. This general-purpose text indexing engine, when applied to music metadata, allows tagging applications to perform metadata matching based solely on the incomplete metadata found in the source music track and a separate index file. MusicBrainz used to match a song to its online metadata by using acoustic fingerprints that attempt to identify music on the actual audio characteristics, but this was slow and inaccurate. These new tools are much better but do have one downside: they require very large index files to run, which delays the instant gratification of tagging your music collection right now, because you first have to download the index files.

Go to the pimpmytunes site at http://musicbrainz.org/wd/PimpMyTunes)to get started. Download the program and the index files pimpmytunes references. Follow the provided instructions for installing the program and its prerequisite software and for where to place the index files.

2.20.1. Configure pimpmytunes

Once you install the application and index files, run pimpmytunes (command line pmt) without command-line arguments to print out the usage for the program and write the default settings file to ~/.pimpmytunes.conf. Using your favorite text editor, edit this file and tweak the default settings to your taste. Pay close attention to the settings listed in Table 2-8.

Table 2-8. Important settings in pimpmytunes.conf




The location of the MusicBrainz Lucene index.


The file naming mask that specifies how to rename tracks (explained later).


The file mask that specifies how to rename various artist (compilation) tracks.


The directory where to save cleanly tagged tracks.


Move files to the destination directory when saving.


Rename files when saving.

pimpmytunes works best when you move cleaned tracks from your music directory to the destination directory. To organize files in place, either turn off the MoveFiles option or set the destination directory to be the same dir as your existing music directory. In the latter case, pimpmytunes will create subdirectories according to your selected FileMask.

Since no two people name their music tracks the same way, pimpmytunes supports file masks, which let you tell the program how to save your files (see Table 2-9). For example, the following file mask:

 %abc/%sortname/%album/%0num - %track 

will instruct pimpmytunes to save the track "Taxman from The Beatles" to the following file:

 /mnt/mp3/clean/B/Beatles, The/Revolver/01 - Taxman.mp3 

Table 2-9. pimpmytunes file masks

File mask



The first letter of the artist name


The name used to sort the artist


Album name


Track number padded with zeroes


Track name

Well, I'm assuming you have set DestDir to /mnt/mp3/cleanwhich you don't have to do. Having a / in the file mask causes pimpmytunes to create a subdirectory, so that all your music gets cleanly organized on your hard drive. Please refer to the pimpmytunes documentation for a full listing of the supported substitution variables.

2.20.2. Run pimpmytunes

Now that you've configured pimpmytunes, you can invoke it with the command pmt, a mode (I'll explain modes shortly), and music files and/or directories containing music files on the command line. pimpmytunes will start up and examine the passed files and recursively search for MP3, FLAC, Ogg Vorbis, and WAV files in the directories.

pimpmytunes operates in two modes: automatic and manual. In the automatic mode (command line option a), the program will check each file and directory passed on the command line to see if it can match the file against the index. If the application finds a match and its confidence rating exceeds the AutomaticThreshold setting, it automatically accepts the match, saves the new metadata to the file, and moves on to the next file. If pimpmytunes can't find a suitable match, it ignores the file and moves on to the next.

Matching files with the automatic mode allows the user to get the mostly clean files out of the way quickly and focus on the troublesome files that pimpmytunes can't readily identify. pimpmytunes in automatic mode can also act in the background to automatically tag files, without user interaction. Setting up a cron job to invoke pimpmytunes periodically allows you to automatically tag files in a download directory. If you download files with mostly clean metadata, pimpmytunes will identify, save, and rename your newly downloaded files automaticallyall according to your preferences.

Operating pimpmytunes in manual mode (-m) presents the top matches found in the Lucene index for each of the files. You need to inspect the matches and enter the number of the match that fits best. Once the match is selected, pimpmytunes saves the file and moves on to the next file. If pimpmytunes didn't find a close enough match, enter e at the command prompt to edit the metadata for this track. pimpmytunes will show the metadata that this track has so far and allow you to edit the information so it can attempt another lookup.

In most cases where pimpmytunes cannot find a match, the metadata tags contained incomplete or incorrect information. For these cases, fixing the data in the tracks with the e command lets pimpmytunes find the right track.

Sometimes MusicBrainz does not have the album that matches your track in that case enter l to open a web browser and look up the artist/track at MusicBrainz. Use the MusicBrainz web site to locate the right album or add a new album if the correct one isn't found. Once you locate the album, copy the album page URL that contains the album UUID (e.g., 8f468f36-8c7e-4fc1-9166-50664d267127), and then switch back to pimpmytunes. In pimpmytunes type a (for album id) and then paste the UUID and hit Enter. This causes pimpmytunes to retrieve the album from MusicBrainz and match the current track to that album.

If pimpmytunes can't find the right match for a track, you can try one of the options from Table 2-10.

Table 2-10. pimpmytunes options




Look up this track at MusicBrainz.


Enter a specific album ID for this track.


Skip this track and try again after processing other tracks.


Delete this track from disk permanently.


Remove this track from pimpmytunes.

If you have trouble with pimpmytunes, go to the MusicBrainz Contact Us page at http://musicbrainz.org/wd/ContactUs and choose your favorite method of getting help.

Robert Kaye

Linux Multimedia Hacks
Linux Multimedia Hacks
ISBN: 596100760
Year: 2005
Pages: 156

flylib.com © 2008-2017.
If you may any questions please contact us: flylib@qtcs.net